Animals and humans collect and process information from their environment using a host of sensory modalities to maintain control of movement. Such integration of information is exceedingly challenging for the control of flight in insects, which is inherently unstable. We use insects as a model system for understanding how multiple sensory inputs influence flight control. While vision is necessary for flight, its processing time is too slow for effective fight control. Thus insects use a faster sensory modality that encodes mechanical information (inertial forces). We use a combination of experimental and theoretical approaches to reveal the precision, sensitivity, and rapid processing speeds of inertial sensors in flying insects. In particular, we suggest that wings not only serve a role as propuslors, but by virtue of the mechanosensory neurons in them, provide a key sensory role for inertial forces. We hypothesize that the wing mechanoreceptors of the hawkmoth (Manduca sexta) rapidly and precisely extract mechanical information necessary for flight control. We further hypothesize that such encoding properties are similar to those in wing evolutionary predecessors, halteres (a sensory organ that is known to be a gyroscopic organ) in flies. To test our hypotheses, we recorded the electrical activity of wing neurons subject to a mechanical stimulus. We used methods from computational neuroscience to characterize the encoding properties (the stimulus features that drive neural responses). We found that nearly all neurons respond rapidly with 5 ms precision and they are extremely selective for the temporal pattern of the stimulus. These results are similar to those found for halteres, and suggest a novel finding that wings serve the dual function of both actuation and sensing. These results reveal a new interpretation of movement control in animals and can serve as inspiration for the next generation of flight systems.

 Glioblastoma multiforme (GBM) is the most common and deadliest cancer of the central nervous system. The existence of stem-like cells within GBM tumors - Glioblastoma stem cells (GSCs) - may explain some of the malignant features of Glioblastoma multiforme (GBM) owing to their rapid self renewal and diffuse migratory behavior. Increasingly apparent is the role ion channels play in many cancer cell properties, including regulation of cell cycle progression and proliferation, tumor cell motility and invasion, and regulation of metabolic rate. We hypothesize that ion channel dysregulation contributes to GSC malignancy, including increased viability, proliferation, and migration. To test this, multiple GSC isolates derived from human GBM tumors were maintained in culture over many passages while retaining stem cell properties. RNA-seq expression profiles revealed ion channel families that were highly enriched in GSC samples. Real-time qPCR was next used to confirm increased expression of voltage gated calcium, sodium, and potassium channels, ionotropic glutamate channels and inwardly-rectifying potassium channels. Identifying ion channels uniquely enriched in GSCs led to functional analyses to determine how ion channel blockers affected their proliferative and migratory properties. We performed MTT cell viability assays to assess how ion channel blockers altered the viability of GSCs compared to a control neural stem cell line. We were able to show a significant decrease in cell viability across GSC lines compared to control at a variety of drug concentrations. Utilizing live-image cell tracking, we determined decreased migratory behavior and rates of these GSC lines with the same blockers. Finally, to elucidate the role of the extrinsic tumour microenvironment on the proliferation and migration of GSCs, unique coculture models involving GSCs cultured on mouse organotypic brain slices were utilized, and showed the affects of stimulating and inhibitory DREADD receptors to modulate local neuronal activity.

As computational processing speeds and algorithmic models for machine learning continue to advance, depictions of Artificial Intelligence (A.I.) in science fiction (SF) have accordingly increased in production. Such fictional narratives typically anthropomorphize A.I., which brings into question whether intrinsically human concepts, such as gender, can legitimately be mapped onto A.I in the direct manner said fictions tend to portray. If “true” A.I., defined here as any technology that accurately replicates a human consciousness, is created, but gender is a social construction derived from strictly biological sex differences, then can the idea of “gender” be said to maintain any relevance to artificially intelligent beings? This paper intends to investigate that question by first cross-examining convolutional neural network theory with models of consciousness from M.I.T. theorists to establish a working model of how a true A.I. could learn, and then using said model, alongside feminist SF critical theory, as a lens through which to compare several SF works which depict gendered A.I.s. Understanding the relation—or lack thereof—between the human concept of gender and A.I.s is highly relevant to the ongoing discussion concerning the rights of A.I.s as intelligent non-humans.

Therapeutic gardens are a specific type of landscape designed to promote physical and psychological human health within a natural environment. Designs of therapeutic gardens vary depending on the user populations and the goal of my project is to understand the healing components of therapeutic gardens and apply them to specific circumstances. In this project, I will conduct a post evaluation on the Ichi-Go Ichi-E Garden, used primarily by Japanese American senior residents of a senior residential community, Nikkei Manor, and analyze the use patterns of residents and staff using observation, mapping, surveys and interviews. By analyzing the success of the project, I hope to answer the question of how therapeutic gardens can be effectively designed for specific users, site conditions and programs. These findings and knowledge will be integrated into the design for a new project: a capstone design & build project led by Professor Daniel Winterbottom for the Puget Sound Veteran’s Affairs Hospital. The garden will serve veterans suffering from post-traumatic stress disorder (PTSD) and other stress-related illnesses. Finally, after gaining feedback from clients and presenting the final design, my classmates and I will build the project by June 2016 when the garden will be dedicated.

Global Positioning System (GPS) technology has revolutionized the way we navigate our world, yet there remain many practical geolocalization problems for which the technology is not directly useful, such as in urban environments and scientific field work where GPS satellite L.O.S. (line-of-sight) obstructions are ubiquitous. Motivated by our collaborations with the LEAF School of Anthropology, our work is focused on providing technological solutions for cm-scale GPS-like geolocalization; particularly for researchers doing field work in prohibitive and complicated field topologies through the use of field-configurable ad-hoc geolocalization subnetworks. We present our work on and our methods to utilize time-of-flight measurements from GPS-enabled acoustic and radio frequency sources placed at the boundaries of the field of interest to determine the geographical coordinates of any number of locations within that field.

The principal technological problem of the renewable energy industry is how to supply energy at its rate of demand and at a marginal cost less than that of present energy storage systems. One of the greatest challenges of this problem is the necessity of storing a portion of total produced energy such that it can be used when the instantaneous rate of energy demand differs from the limits of its instantaneous rate of production to balance out the demand and supply for energy. Our research focuses on the storage problem of current Under Water Compressed Air Energy Storage (UW-CAES) systems; in particular on decreasing the costs and environmental impact of air storage by performing electrolysis in a deep-sea electrolyzer to capitalize on the extreme energy densities already-available at depth and increase the energy density of each storage vessel/balloon by storing both hydrogen and oxygen. We demonstrate a method that increases the efficiency of energy storage by performing electrolysis using the high pressures available at depth, thereby eliminating the use of costly compressors and create a condition for the apparatus to be as close to a thermo neutral state.

The ability of animals to detect minute chemical concentrations and locate odor sources is unparalleled in synthetic systems. This is especially true in the hawkmoth Manduca sexta which can track and find the source of chemical plumes with astonishing efficiency. Male moths, for example, can detect female pheromones at concentrations of 1 in 10¹⁷ molecules of air. In this work, I created a biological odor sensor by measuring the electrical activity of the neurons in a severed antenna in response to their odor detection, (a.k.a. an electroantennogram or EAG). By comparison, commercial gas sensors are incredibly inefficient, large, heavy, expensive, and have slow response times. I have developed a small quadrotor drone which uses an electroantennogram with the antenna of a Manduca sexta moth to track a chemical plume to its source. The EAG signal is measured as the voltage across the antenna and then amplified and filtered to reduce noise. The control module of the drone uses this signal in conjunction with a measurement of the ambient airflow to steer the vehicle. The drone uses a cast and surge search algorithm to follow the chemical plume upwind: flying crosswind until the chemical is sensed, flying upwind until the scent is lost, and then repeating the process. This experiment provides proof of concept that EAGs can be used to direct the flight of an aerial vehicle for the purpose of chemical localization. This provides an odor sensing platform that is many orders of magnitude lighter than commercial odor sensors. The applications of a small, reproducible, chemosensing drone are numerous, ranging from locating contraband to testing biological hypotheses of insect flight.

Factors such as temperature, pressure, presence of a catalyst, and reactant concentration are all known to affect the rate of chemical reactions and are well exploited in industry to drive chemical synthesis to favor certain mechanistic pathways. Here we consider the effect of using magnetism to control the rate of reaction for the Briggs Ruascher reaction. This reaction was chosen as a model for magnetic manipulation because it is mediated by both radical and non-radical pathways that compete with oscillatory behavior to produce the same product. Dramatic color changes serve as a visual cue to indicate which path is dominating. Here we put forth magnetic radical arrest hypothesis for the Brigg’s Ruascher reaction. It is known that electrons act like tiny magnets that pair with opposite spins. Because radical species carry a lone unpaired electron, presumably their magnetic fields are not concealed. We proposed that due to the exposed nature of the unpaired radical electrons in the Briggs Ruascher reaction, the progress of the reaction would be affected by the application of a magnetic field. Upon exposure to strong magnets, the radical intermediates of the Briggs Ruascher reaction were expected to aligned their spins with the applied magnetic field affecting their reactivity. Several types of neodymium magnets with various strengths were applied to the reaction in order to determine whether magnetism could prevent or greatly reduce the velocity of the color phase changes for the radical mediated portion of the reaction. We also built an electromagnet to expose the reaction to in order to achieve even greater field strength. The Brigg’s Rouscher reaction was found to be significantly robust and there was insufficient evidence to suggest that the reaction was affected by neodymium fields below a range of 5.8 mT. Results of exposure to the electromagnet are still in progress. The color changing cycles of the Briggs Ruascher reaction were found to follow a decreasing sigmoid mathematical model before going to completion after 9-10 oscillations.